Scientists hope to glimpse
the secret life of clouds
Posted June 12, 2009
(Click to open video in a separate window.)
Colorado State’s David Randall says global climate models must adequately account for clouds. The forecast calls for computational muscle.
Despite decades of analyzing satellite and simulation data, scientists have barely begun to predict clouds’ behavior or to understand their major influence on the atmosphere and global climate.
Many questions remain. For example, how do cloud particles form, interact and precipitate or re-evaporate? What determines the number of clouds in a given area? How do clouds interact with turbulence and visible and infrared light?
David Randall, an atmospheric scientist at Colorado State University, leads a team seeking answers to those questions and others with a new global cloud-resolving model (GCRM) that will use computers to portray Earth’s atmosphere in 10-second snapshots.
The GCRM, part of DOE’s Scientific Discovery through Advanced Computing (SciDAC) program, is built on a geodesic grid that consists of about 100 million mostly hexagonal columns, each with 128 levels. The 128 layers will cover a layer 50 kilometers up from earth. For each of these grid cells, the model predicts the wind, temperature and humidity at points just 4 kilometers (and eventually 2 kilometers) apart. That’s an unprecedented resolution; most global atmospheric models provide detail at a 100-kilometers scale.
“No one has done this before in quite this manner, and it’s our hope that our project will point the way to future generations of models,” says Randall, a member of the Intergovernmental Panel on Climate Change (IPCC) science team that shared the 2007 Nobel Peace Prize with former U.S. Vice President Al Gore. Randall was a coordinating lead author of a chapter about climate model evaluation in the IPCC’s final report.
Computer speed is of the essence. The model is using 80,000 processors on Jaguar, the Cray XT supercomputer at Oak Ridge National Laboratory (ORNL). Even with that much computational punch, the GCRM can complete only a few simulated days per wall-clock day of calculation. Data archival speed and capacity also are critical. A run of just a few simulated days can easily produce several petabytes (quadrillion bytes) of output.
Randall and his colleagues succeeded in a brief test of the GCRM on computers at the National Energy Research Scientific Computing Center (NERSC) at Lawrence Berkeley National Laboratory. “Almost all of the development work has been done at NERSC,” he adds.
In the next couple of months the team will aim at capturing two simulated days, with a goal of running and archiving an entire year before the project ends in 2011. Simulation of an annual cycle will be a landmark, Randall says, because a year represents the basic unit of climate.
“There are three things we’ve wrestled with in building the GCRM,” Randall says. “One is which equations we should be using. The second is finding the right methods to solve the equations. And the third is getting a computer to actually compute the solution fast enough to be useful.”